External Validation of A Decision Tree Model for Predicting Cancer-Specific Mortality After Radical Cystectomy: A...

Area of Science:

• Urology
• Oncology
• Biostatistics

Background:

• Prognostic tools are vital for urothelial bladder cancer (UBC) management post-radical cystectomy.
• Decision tree models show potential but require external validation for clinical utility.
• This study validates a predictive model in a Spanish multi-centre cohort.

Purpose of the Study:

• To externally validate a decision tree model for predicting cancer-specific mortality.
• To assess the model's performance in an independent Spanish cohort.
• To determine the clinical utility of the validated model.

Main Methods:

• Multi-centre retrospective cohort study of 553 patients undergoing radical cystectomy.
• External validation of a decision tree model developed using SEER data.
• Analysis included concordance index (C-index) and decision curve analysis (DCA).

Main Results:

• The model achieved a C-index of 0.733, indicating good discriminative ability.
• Decision curve analysis showed clinical utility at threshold probabilities >25%.
• The model maintained predictive accuracy despite cohort differences.

Conclusions:

• The decision tree model is externally validated for predicting cancer-specific mortality post-radical cystectomy.
• The model demonstrates robust performance in the Spanish cohort.
• Findings support the model's application in clinical practice in south-east Spain.